Specific embodiment
In this detailed description, Primary Reference is begged for the circular symmetry optical element of total internal reflection (TIR) collimator form
By the various embodiments of optical element of the invention.It should be noted that this is not limited the scope of the invention, the scope is same
Suitable for the optical element of refraction or reflective collimator form.
Fig. 1 schematically illustrates the layout of illumination component 100, and it includes the hair covered by wavelength converting phosphors element 104
Optical diode (LED) chip 102.Two different ray paths 106 and 108 are shown, it corresponds to different angles from phosphorescence
The light that the surface of volume elements part 104 sends.The angle is defined so that being sent on the direction perpendicular to the surface 110 of phosphor elements
Light be defined as 0 °.Therefore, the light for being sent parallel to the surface is defined as 90 ° and sends angle.Send angle t quilts
It is defined as sending the angle between the normal on light and surface 110.Due to passing through phosphor elements in a direction orthogonal to the surface
The optical path of 104 light 106 is shorter than the path of the light 108 with more high angle, so there is change of the color with angle.
In fig. 2, curve map 200 is depicted, its x for being shown as the function for sending angle t, y colors point.For given
LED 102 and phosphor elements 104 are combined, and measurement has been shown that the light sent by phosphor LED combination is in x, the y colour spaces
Straight line 202.Here, LED is blue led.At t=0 °, light for nattier blue, and t=90 ° at, light be it is flaxen,
In other words as shown in Figure 2, the x and y coordinates at low-angle are less than at wide-angle.The average color point of LED can be with
Measured with integrating sphere, it is indicated by the cross 204 on line 202.Accordingly, there exist being defined as tAAngle, at the angle,
The color point of the light sent by LED is equal to the average color point of LED.For the color according to Fig. 2 with the change of angle, may incite somebody to action
Light be collimated to its the equal color point in angled place a wave beam in.Minimum requirements is that optical element must have
There are at least two sections.This two sections (mirror) or its combination that can be to reflect (lens) or reflection.
Fig. 3 is schematically illustrated and implemented with the example of the optical element 300 of circular symmetry TIR collimator forms according to the present invention
The cross section of mode.For example, the collimator can be made up of transparent plastic material.Optical axis 301 is defined as in optical element 300
Heart axle, it is arranged to the central axial alignment with phosphor elements 100.Additionally, the root edge 300 of optical element is disposed in and phosphorus
In the identical plane of surface 110 of body of light element.Optical element 300 includes the first interior section 302 and the second exterior section
304.With small angle, i.e., the light that small t sends is collimated by the interior section 302 of the form of lens at the middle part of optical element 300
To the gained light beam that the substantially flat light-emitting area 310 from optical element sends.The light sent with big angle is by outside portion
Points 304 side surface is received, and after this, it is reflected in curved surface 308, and towards the light-emitting area 310 of optical element 300
Redirect.Curved surface 308 is not configured to reflected light must be coated by mirror material, because light will be incident with sufficiently large angle,
So that it will be reflected by total internal reflection.
Optical element 300 is designed such that optical axis 301 and impinges upon the mistake between interior section 302 and exterior section 304
The angle crossed between light ray a little is equal to angle tA.By by the optical surface of collimator by angle tAGiven point punishment
It is two sections, can as follows designs the curved surface 303 and reflecting surface 308 of lens component 302:For from light-emitting area 310
Each sends angle, and color point is identical.So as to CoA changes can be reduced or eliminated, so as to provide homogeneous in far field
Light output.
The mathematical theory for describing the design of the curved surface of optical element is the work [Mathematical based on M.Maes
Method for Reflector design,M.Maes,PhD thesis 1997,ISBN 90-74445-36-5].According to upper
The work of face reference, it can be shown that curve form can be calculated by solving two couples differential equations:
Wherein N is the number of section, and I is the intensity distribution of light source,For the target strength of collimated wave beam is distributed, x is x CIE
Color point coordinates, and d for divergent beam be 1 and for converging beam be -1.Function η describes source and sends angle t and standard
The relation between angle, θ in straight wave beam.The function couples acquisition by solving coupled differential.This function is used, for example
Can be with the shape of calculating optical element.In " the An inverse method for the design of of C.R.Prins et al.
Provided in TIR collimators to achieve a uniform color light beam " and draw above-mentioned equation
The further exemplary description of mathematical derivation, it is published submitting to and is entirely incorporated into this by reference, and it is attached
At the ending of this application of its composition part.
General explanation is that optical element is configured such that 0 ° and tABetween each angle be coupled to from optical element
It is predetermined send angle, and for 0 ° and tABetween each angle, exist and be coupled to from the same predetermined of optical element
Send angle in tAAnd the corresponding angle between 90 ° so that angle is sent for all, the color point for sending light is identical.By
Angle t differences are sent for different in the flux of the light sent from phosphor elements, so it is determined that the song of optical element 300
During the design in face, it is considered to Flux Distribution.Clashed into by ray in the amount of the flux from each light ray direction of optical element 300
The local surface curvature control of lens and reflector.Accordingly, it is possible to realize different Flux Distributions.For example, it is desirable to distribution can
Being the Gaussian beams with 35 °F of WHM (full width at half maximum) angles.Based on above-mentioned mathematical method, easily realize providing Gauss or clock
The optical element of shape intensity distribution.
Fig. 4 illustrates the x as the function for sending angle t combined for LED and phosphor elements, y colors point, wherein
It is not linear function that CoA changes in x-y space.In fig. 4, CoA changes are described by parabola 402, and it can be mathematically
Represented by quadratic polynomial.Non-linear CoA changes can occur when LED is combined with complicated phosphor system.For example containing
The phosphor system of more than one phosphor layer or including scattering particles, in the preferred direction scattering light (photon effect) phosphorescence
In volume elements part.
In order to realize the homogeneous light output for changing according to the CoA of Fig. 4, it is necessary to using as shown in Figure 5 including three
The optical element 500 of individual part 502,504 and 506.The friendship determined between the one 502 and the 2nd 504 part can be found from Fig. 4
The angle t of crunode1And the angle t in the crosspoint between determining the 2nd 504 and the 3rd 506 part2.The average color point of LED is by song
Cross 404 in line chart 400 is represented.Straight line draws from t=90 °, by average color point 404 and continues forward intersecting
CoA change curves 402, so as to find the angle t of first paragraph in intersection1.By the program being similar to, by straight line from t=0 °
Draw, by tA, and to curve 402, it provides second angle t2.Therefore, as shown in Figure 5, optical element 500 is according to t1
And t2It is divided into three parts.The principle of operation is identical with the optical element 300 with two parts, except three coupled differentials
Equation must be solved, to calculate the curvature of curved surface 508,510 and 512.However, it is possible to use identical mathematical theory.Should
Note, what the precise shapes of the curvature of part also depended on that designer makes send, and light beam will assemble or the selection of diverging.
In principle, the optical element of the CoA changes for being described by higher order polynomial can be with similar to as described above
Based on basic mathematical description mode design.
Even if the present invention is described by reference to its specific embodiment, but many different changes, modification etc. will
Those skilled in the art are become apparent.For example, change due to the CoA of light source can be measured, so need not be in theory
The wavelength and Flux Distribution of light source are described.It is thus possible to be used together according to this with substantially having any light source that CoA changes
The optical element of invention.
In addition, technical staff is in invention required for protection is put into practice, can be to accompanying drawing, disclosure and appended right
It is required that research in, understand and realize the change to the disclosed embodiments.In the claims, word " including " be not excluded for it
His element or step, and indefinite article "a" or "an" is not excluded for multiple.Only in mutually different dependent claims
Described in certain measures the fact that, not indicating that the combination of these measures can not be advantageously used.
For designing TIR collimaters to realize the inverse approach of uniform color light beam
C.R.Prins §and J.H.M.ten Thije Boonkkamp§
Eindhoven University of Technology
T.W.Tukker and W.L.IJzerman
Philips Research, Philips Lighting
Summary:Color in the light output of White LED is common and an open question with angle (CoA) change.This
Wen Zhong, we introduce new method, change with by using special collimater reduction CoA.The method is based on the inverse design of parsing
Method.We provide the numerical algorithm for solving the differential equation for coming from the method, and use Monte Carlo (Monte-Carlo)
Ray tracing demonstrates result.
Keyword:Optical design, LED, inverse approach, optics, Functional Design method, color with angle, color homogeneity,
TIR collimaters, weighted color mixing, the blend of colors based on colourity
1. introduction
White LED technology is in the biography for surmounting such as compact fluorescent lamp etc in terms of light output, light quality and efficiency
The node of light technology of uniting.Since the seventies in last century, the flux output every ten years per LED illumination device increase for 20 times with
On, and the production cost per lumen drops to 1/10th [4] in every ten years.LED does not sell individually, and on the contrary, they are configured to
Lighting apparatus including one or more LED, optical system, electronic component, fin and beautiful shell.Optical system is often
Focus light at the collimater on specific direction.Herein, it is contemplated that single led and collimater combination.
______
Department of Applied Physics
§Department of Mathematics and Computer Science
2012Kluwer Academic Publishers.Printed in the Netherlands.
Unfortunately, it is difficult to create the LED of light of the transmitting with flat-white.Create in the following way such
LED:Blue led is set to be coated with the layer and the possibly extra play of red-emitting phosphor of yellow phosphor.This phosphor coating
Some blue light is converted into gold-tinted or feux rouges, so as to produce white light.The ratio of the light changed depends on light to pass through phosphor
The distance of layer, and this is relevant with the angle of departure of light.Therefore, the color of launching light is related angle:Perpendicular to surface emitting
The inclined Bluish white of light, and the inclined yellow of light [10, p.353-357] for being nearly parallel to surface emitting.This is referred to as color with angle (CoA)
Change.
Numerous studies have been carried out to change with by changing this CoA of LED reductions.Draw often through in phosphor layer
Enter bubble [14] or CoA changes are reduced by application dichroic coating [9], but these methods also reduce efficiency and increase
Plus the production cost of LED.If LED is used with collimating optic combination, can be by using micro- at the top of the collimater
Structure is seen to reduce CoA changes.This is a kind of technology being widely adopted.However, generating process of the microstructure in collimater
It is middle to introduce extra cost, and collimater is appeared not to attraction, and widen light beam.
A kind of alternative is introduced herein, is changed with using free-form collimater reduction CoA.Using free-form
Collimater reduction CoA changes have two advantages.First, the special optical element will not introduce extra light loss, and repair
Changing LED would generally introduce extra light loss.Secondly, expensive LED modifications are no longer necessary, so as to cause whole optical system
The cost reduction of system.Most of color mixing approach by mixing the light of many different angles from light source based on reducing face
Color change.Wang et al. [11] have studied and be changed using dome reduction CoA of the mixing from two light of different angles.They refer to
To go out remove completely in theory CoA changes, but they are not shown proof.Herein, we are introduced into inverse based on parsing
Method for designing [2,6,7], its really possible constructive method for showing removal CoA changes completely.
How Section 2 description designs TIR (total internal reflection) collimater for having and specifying output intensity to be distributed using inverse approach,
And Section 3 illustrates how to include blend of colors in the method.Then, Section 4 discusses numerical computations and checking.Finally,
Conclusion is given in Section 5.
2. TIR collimaters are designed using inverse approach
Before blend of colors is discussed, we will illustrate how to use inverse approach to design TIR collimaters.TIR collimaters are
Rotationally symmetrical collimater, generally as the transparent plastic system as makrolon (PC) or polymethyl methacrylate (PMMA)
Into.In Fig. 1 it can be seen that the profile of TIR collimaters.
Design program includes two steps:First, it would be desirable to find the angle t and light for leaving source in light and leave standard
Relation between the angle, θ of straight device, so-called transmission function.Then, we transfer function to calculate TIR collimaters using these
Free Surface A and C.
The profile of Fig. 1 .TIR collimaters.Complete TIR collimaters can be obtained by making profile be rotated around x-axis.Spot light
S is located at origin.Collimater includes four entirely different surfaces:A, B, C and D.Solid black lines indicate flat fixation surface, dotted line
It is Free Surface to be calculated.All light in collimater are all reflected or reflected once by Free Surface just.
2.1. transmission function
Transmission function η:[θmin,θmax]→[tmin,tmax] the angle t ∈ [t that describe in the light launched from sourcemin,tmax]
With the angle, θ ∈ [θ of the light launched from TIR collimatersmin,θmax] between relation.The radiation direction of light is by it relative to x-axis
Angle counterclockwise represents.Interval [tmin,tmax] can be the source angle of departure subset.Note, we can also select the transmission function to be
[tmin,tmax] to [θmin,θmax] mapping.The latter's selection is more conventional in the literature, however, as will be seen that behind us, I
Selection be more suitable for the purpose of blend of colors.It is strictly monotone and so that reversible that we select transmission function.
We find transmission function using Flux conservation.Assuming that rotationally symmetrical intensity of the I (t, u) [lm/sr] for light source
Distribution.Here t ∈ [0, pi/2] be relative to the angle of symmetry axis (inclination angle), and u ∈ [0,2 π) be the angle rotated around symmetry axis
Degree (azimuth).Because the symmetry of system, intensity I (t, u) is unrelated with u, and is represented by I (t).We are by making I (t) right
Angle u integrates to introduce active strengthIt is the every radian divided by 2 π by the endless belt [t, t+dt] on unit ball
(rad) flux:
The unit of active strength is [lm/rad].Similarly, there is the intensity distribution of the light from collimater outgoing, by G (θ,
φ) expression, wherein θ ∈ [0, π] and φ ∈ [0,2 π) it is inclination angle and azimuth.We require that the target distribution is also rotationally symmetrical
, and we represent it with G (θ).Integrated by angle φ, it has been found that effective target intensity distributionTransmission function should cause that the intensity of the light for leaving collimater has required intensityMore detailed description to active strength distribution can be found [6] in the achievement in research of Maes.
Be necessarily equal to from source transmitting in η (θ) from the luminous flux (unit lm) between θ and θ+d θ of collimater transmitting and
Luminous flux between η (θ+d θ).This produces following relation:
Wherein, for the transmission function of monotone decreasing, σ=- 1, and for the transmission function of monotonic increase, σ=1.It is logical
Cross and (2) are transformed into the differential equation, transmission function can be calculated:
It can be in initial value η (θmin)=tmin(for increasing η) or η (θmin)=tmax(for reduce η) situation
Lower integral.
We have to make sure that for the Flux conservation of whole optical system.(2) from θminTo θmaxIntegration is produced with ShiShimonoseki
System:
Pairing functions are made in requirement (4) to the luminous flux in systemLimitation.If be unsatisfactory for, system does not have
Physical significance.
2.2. for the TIR collimator designs of known transmission function
For some tav∈ [0, pi/2], it is [0, t that we have scopeav] and [tav, pi/2] two transmission functions
When, we can design TIR collimaters.As shown in figure 1, TIR collimaters include surface A until D.Light passes through two potential routes
Propagate through collimater.In first route, light is reflected by surface B, reflected by total internal reflection by surface C and finally
Reflected by surface D.In scope [tav, pi/2] in angle under will comply with the route from the light of LED emission.In Article 2 route,
Light is reflected by surface A and then reflected again once by surface D.In scope [0, tav] in transmitting light will comply with the route.
Given transmission function η, we can calculate the angle before being reflected by surface D in angle t and lightBetween
Relation.By snell law (Snell ' s law), we obtain relationWherein n is collimater
The refractive index of material.This provides relation
The calculating of Fig. 2 Free Surfaces C.This figure is the amplification of a part of Fig. 1.The path of light is represented by gray line.
First, it is considered to the light of pass course S-B-C-D.The light is reflected at the B of surface, rolled over by C reflections and finally by D
Penetrate.Assuming that d is from the left end of surface B to the distance in source, and α is clockwise angle of the surface relative to x-axis (see Fig. 2).
Assuming that (xB(t),yB(t)) for surface B profile coordinate.Our selection parameters so that the light in source is left under angle t
Will be in (xB(t),yB(t)) place's hit surface B, and the surface is left under angle s (t) relative to symmetry axis.We use
Basic geometry is derived:
Wherein (6c) is derived using the snell law of refraction.We can use what is developed by Bortz and Shatz now
' generalized function method ' gauging surface C [2,3].They have derived the following differential equation, and its variable is illustrated in fig. 2:
Here, f is the distance that light advances to surface C from surface B, and q (is defined at t=pi/2s along the arc length of surface B
For 0), δ be relative to launch point at normal measured counterclockwise the light angle of departure, and s be light leave surface B relative to
The angle of symmetry axis.It is reflection or refraction that angle beta depends on optical surface.For reflecting surface, have [2]
Generally it is understood that q and s are the function from the angle of departure t of original light source.Make (7) to be multiplied with dq/dt to be given:
From (6a) to (6c) we derive
We can calculate all functions of our needs in the hope of solving equation (9) now.By relation (5), (6c), (8),
(10a) and (10b) substitutes into equation (9), and calculates function f.The location expression of reflecting surface is
xC(t)=xB(t)+f (t) cos (s (t)), (11a)
yC(t)=yB(t)+f (t) sin (s (t)), (11b)
The light of pass course S-A-D is considered now.Similarly, we can be with the position of gauging surface A.Because being incident on A
Light directly from spot light S, we can take q (t)=0 and s (t)=t.Which greatly simplifies (9).For β, refraction is set
With following formula [2]:
IfThen from angle t to angleRefraction be physically impossible.By q
T (), s (t) and β substitute into (9) and produce equation:
And we obtain the following description to surface A:
xA(t)=f (t) cos (t), (14a)
yA(t)=f (t) sin (t) (14b)
3. color-weighted TIR collimaters
In Section 2, we have seen that how to calculate the freedom of TIR collimaters for given source and target intensity
Surface.In this section, we show how color homogeneity is incorporated into this model.We introduce some of people's perception color first
Theory, and derive two equation groups of the ODE similar to equation (3) of coupling.Then, we discuss the equation
The removable singular point occurred in group.
3.1. it is used for the transmission function of blend of colors
Perception on people to color has been carried out substantial amounts of research [8,15].It has been found that light beam can be led to by its light
(unit lm) and two dimensionless chromaticity coordinate x and y with value between zero and one are measured to fully describe.Chromaticity coordinate can
In being calculated used in blend of colors.Assuming that we have two-beam, numbering is 1 and 2 respectively, with luminous flux L1And L2And color
Coordinate (x1,y1) and (x2,y2).In the achievement in research of Malacara [8, p.57-58,103-105], we can be found that two
Plant the chromaticity coordinate (x of the light beam of colorT,yT) be changed into after blending:
The scatter diagram of the x and y chromaticity coordinates of the measurement of the LED used in numerical experiment Fig. 3 herein.Value is substantially
On straight line.Round size corresponds to active strength.Measurement data in the lower left corner corresponds to perpendicular to LED surface transmitting
Light, the measurement data in the upper right corner corresponds to the light parallel to LED surface transmitting.
Note, resulting chromaticity coordinate (xT,yT) in (x1,y1) and (x2,y2) between straightway on.For infinitesimal
Angle dt1And dt2, in t1And t1+dt1Between and in t2And t2+dt2Between luminous flux be respectivelyWithRule (15) is converted into
From measurement data, it has been found that the linear approximate relationship between x (t) and y (t), so as to mean all colourities
Coordinate (x (t), y (t)) is all on straight line (see Fig. 3).Therefore herein we assume that linear relationship.We are also it has been observed that remove
(see Fig. 3 and Fig. 4 b) outside the region being practically free of at the wide-angle of luminous flux, x (t) and y (t) is the increasing function of t.
For simplicity, we it will be assumed that x (t) and y (t) are strictly increasing.
LED light source is typically lambert's (Lambertian) or close to lambert, that is, the intensity distribution I of the light launched
(t, u) is proportional to cos (t), or with the intensity distribution close to this.Factor sin (t) is multiplied by for rotationally symmetrical setting,
LED typically will be with close Active strength distribution.Exact shape can change, but I
Can assume alwaysRight-hand derivativeAnd left derivativeIt is also assumed that, for all of 0<t<Pi/2,Similarly, Wo Menyou
And right-hand derivativeAdditionally, we will assume that, for all of 0<θ≤θmax,
In order to construct the collimater of the light in mixing source as follows:Color point at target is constant and target at will
The intensity is asked to beWe are by interval [θmin,θmax] it is separated into N number of different sections.Define ordered list 0=τ0<τ1<...<τN=
π/2.Each section of i is subinterval [τi-1,τi]From the angle of departure of light source and from collimation
Relation between the angle of departure of device is by transmission function ηiDefinition:For remote
Each angle, θ at the target of field, will have from a light for angle in lucky each section.Light under the angle at target
Colourity will be the colourity from different sections of light weighted average.According to definition, each transmission function is dull, reversible reflecting
Penetrate.For the ease of mark, we are introduced into following agreement:Be at source (in section i) direct projection to angle, θ
Light intensity.Similarly, for i=1,2 ..., N, we write out xi(θ)=x (ηi(θ)) and yi(θ)=y (ηi(θ))。
Direct projection is to the light of angle, θ by from the N number of different angle t at sourcei=ηi(θ).Similar to (2), we have luminous flux
Conservation
And similar to (16), we have by (xT,yT) represent target colorimetric:
Here, for monotone decreasing ηi, σi=-1;And for monotonic increase ηi, σi=1.We will show (x belowT,
yT) it is weighted average colourity.Chromaticity coordinate (xT,yT) must be on the straight line relevant with x (t) and y (t).From blend of colors rule
(15) it will be seen that must be on the straightway between the colourity of original beam by the colourity that two light beams of mixing are produced.
Then, (18a) or (18b) is unnecessary.It is most convenient to abandon (18a), because (18a) is slightly more more complicated than (18b).From
(17) and (18b) we derive following differential equation group:
Equation (19) describes two equation groups of the ODE of coupling.Hereinafter, we select N=2, because
Otherwise equation group is to owe fixed.
From (18a) and (18b), we can be derived for calculating (xT,yT) expression formula.First by (18a) and (18b) weight
It is written as
The domain of transmission function is adjoined.Therefore, for any function F (s), Wo Menyou
And so as to be integrated generation to (20)
It shows (xT,yT) really weighted average chromaticity coordinate.
Before these relations are assessed, we will have to select σ1And σ2Value.At present, we will select σ1=1 and σ2
=-1.This selection implys that η immediately1And η2Initial value.For some τ1∈ (0, pi/2), function η1Should be by [0, θmax]
It is mapped to interval [0, τ1], and should monotonic increase.This implies η1(0)=0.Similarly, function η2Should be by [0, θmax] reflect
It is mapped to interval [τ1, pi/2], and η2Should monotone decreasing, so η2(0)=pi/2.If coefficient matrix (19) is not unusual
, we can derive η '1With η '2Following formula:
3.2. removable singular point
At θ=0, becauseAndWe obtain being removed with 0 in (23)
0。
Theorem 1. is assumedAndIt is also supposed that η '1(0) ≠ 0 and
η′2(0)≠0.At θ=0, Wo Menyou
Prove.Because previously mentionedη1(θ) and η2The property of (θ), at θ ↓ 0, Wo Menyou
WithWe also haveWe can use L'Hospital rule (l'Rule) calculateWith
Note, as previously defined,AndIt is left derivative and right-hand derivative.For θ ↓ 0
(23) are assessed by substituting into (25) produce (24).It is furthermore noted that because 1/y (t) be monotonic function,AndExpression formula under root is always strict positive.
Lemma 1. is for all of θ ∈ [0, θmax], η1(θ)≤η2(θ)。
Prove.Assuming that for some 0≤θ≤θmax, η1(θ)>η2(θ).The integration of (19a) is produced
By using (26), we have:
(27)
Wherein last equation comes from Flux conservation.This illustrates for η1(θ) and η2These values of (θ) are unlikely to be
The solution of integral equation, and we may safely draw the conclusion η1(θ)≤η2(θ)。
We define t nowavSo that y (tav)=yT.It will be seen that η1(θmax)=η2(θmax)=tav。
The η of lemma 2.1(θmax)=η2(θmax)=tav。
Prove.We deduce η from lemma 11(θmax)≤η2(θmax).From (26) it will be seen that for η1(θmax)<
η2(θmax), it is unsatisfactory for Flux conservation, and we conclude that η1(θmax)=η2(θmax).To (19a) and (19b) from θ to
θmaxIt is integrated to find
By using η1(θmax)=η2(θmax), this is simplified to
Assuming that for some θ, η1(θ)>tav.Because y (t) monotonic increases, for t >=η1(θ), we have yT/y(t)<1,
And it is unsatisfactory for (29).Similarly, if η2(θ)<tav, then it is unsatisfactory for (29).We may safely draw the conclusion η1(θ)≤tavAnd
η2(θ)≥tav.For θ=θmax, with reference to the two results, we may safely draw the conclusion
η1(θmax)=tav=η2(θmax). (30)
Fig. 5 b show two results of lemma:η1Curve always in η2Lower section, and two curves end at tav.This
Cause the second singular point at θ ↑ θ max:η1(θ) and η2Both (θ) all approaches yT, and again in (23) we obtain using 0
Except 0.
Theorem 2. (23) is in θ=θmaxThe singular point at place can be removed using following formula
Prove.By using L'Hospital rule and the chain rule for differential, we derive from (23):
For η1(θmax) and η2(θmax) solve these generations (31).Here we have ignored η '1(θmax)=η '2(θmax)
=0 possibility, but substitute into (19) show ifThen η1And η2Derivative really must be non-zero.
4. numerical result
Method described in Section 2 has been directed to the LED with extra high CoA changes and has been tested.Measure
The colourity and intensity of the LED, and the data are used to calculate the Free Surface of TIR collimaters.Because degree is more often used in optics,
Calculating in this section and experiment are spending rather than radian is expressed.We have evaluated using ray tracking soft LightTools [1]
The performance [1] of TIR collimaters.
4.1.LED modeling
LED is measured using goniophotometer.Goniophotometer is the strong of light of the measurement under different solid angles
The equipment of degree, colourity and many other characteristics.Our LED is that 46 different angle t between 0 degree and 90 degree are (relative
In surface normal) under and measure under 4 different angle u.Because it is assumed that the rotational symmetry of system, is worth to angle
U is averaging.
The measurement intensity of LED uses multinomial t in Matlabi-90i(wherein i=2 ..., 7) it is fitted.
This group of multinomial is selected, because they are zero at t=90 °.Additionally, I (t, u) is smooth, therefore this group of multinomial is in t=
At 0 ° must have relative to t be zero derivative.Linear least squares fit [5] produces coefficient Ci.Following active strength
Function has been used in Matlab calculating:
X and y chromatic values have used multinomial ti(being directed to i=0,2,3,4,5,6,7) is fitted.Select these multinomial
Formula because they at t=0 have be zero derivative.Linear least squares fit produces coefficientWithIn Matlab
The colourity function used in calculating is
Multinomial in using (33), (34a) and (34b) in the solution differential equation (23).Can in Fig. 4 a and
Polynomial figure is seen in 4b.In 3.1 trifles assume monotonicity be unsatisfactory under the angle more than 70 °, but because
Small luminous flux in the range of this, this does not interfere with the method for value solving and solvability for ODE.
In ray tracking program LightTools, build 3D models to simulate LED.Model is come using 46 emitting surfaces
Construction.Light of each surface emitting in the range of certain angle.Surface k=2,3 ..., 45 are transmitted in angular range (2 (k-
1) -1,2 (k-1)+1) the interior, light with intensity corresponding with the measurement result under angle t=2 (k-1) and chromaticity coordinate.Table
Light corresponding with the measurement data under angle 0 and 90 is launched in face 1 and 46, but is only sent out in scope (0,1) and (89,90)
Penetrate.Far field receiver is added in model, the angle of its launched light of measurement and is followed the trail of according to Monte Carlo ray
Simulation calculates intensity and colourity pattern.
In figs 4 a and 4b it can be seen that the measurement data of the LED without collimater, least square fitting and
The comparing of the ray tracing result of LightTools models.Measurement data shows intensity pattern close to lambert, but in wide-angle
Under, show some scramblings in intensity and the aspect of chromatic value two.These scramblings are due to the defect in measurement.Above
The scatter diagram of measured x and y chromaticity coordinates is had been illustrated with figure 3.The figure do show the near-linear between x and y
Relation.
The comparing of the measurement data, least square fitting and LightTools models of Fig. 4 .LED.
4.2. the calculating of transmission function
Example calculations have been carried out by using the data such as seen in figs 4 a and 4b.Target strength selection be
Gaussian Profile [13] with full width at half maximum (FWHM) [12] under 20 °.This produces effective target intensity
Wherein 0≤θ≤1.25 θFWHM, θFWHM=20 °, and select C to cause
By using the ode45 in such as Matlab, we can solve the differential equation (23).For small under small θWithValue, relation (24) should be used.However, for θ → θmax, numerical solution (Fig. 5 a) shows to use (31)
The singular point of solution.At this point, η1(θ) and η2The curve of (θ) approaches line t=tav.Because calculating yT、tavIn and solve
Small calculation error in the differential equation, eta-function will not accurately converge to tav.In an eta-function and line t=tavIntersect
When, the symbol of the derivative of other eta-functions changes.Since the point, error increases rapidly.
The solution of this problem is to realize self-defined ODE solvers, and its each step in θ recalculates yT。
This will not significantly change yTValue, but this it is small amendment significantly stabilize solver.For the experiment, realize self-defined
Runge-Kutta methods.Method is calculated in the discrete levels θ with fixed step size0,θ1,...,θNsUnder η1And η2.Each
Individual θiPlace, y is recalculated according to following formula to still unintegrated angleTValue:
Then, in each step of Runge-Kutta algorithms, valueFor replacing the y in (23)T.This method stabilization
Solver.For Ns=500, the run time of the algorithm is several seconds.In figure 5b it can be seen that for being found
Example problem profit solution in this way.
Fig. 5 using two kinds of difference ODE solvers use fitting data and there is the mesh of full width at half maximum at 20 degree
Mark intensity(23) solution.Note θ=θ in ode45 solversmaxThe unstability at place.
4.3.TIR the assessment of collimater
Then, by using the method described in the 2.2nd trifle, calculate TIR using the transmission function for obtaining and collimate
Device.The collimater is converted into LightTools models.It is 500 points that each Free Surface is discretized.In figure 6 can be with
See the effect of the LightTools models of the collimater.Radius of the opening with 5mm in the bottom of collimater, and it is accurate
Straight utensil has the height of 9.47mm.
The effect of the LightTools models of the TIR collimaters designed by Fig. 6
By following the trail of 10 in non-dispersive model6Light, the model of LED of the assessment with TIR collimaters.In order to ensure point
Source behavior, the size of LED models is reduced for 0.01mm and multiplies 0.01mm.In the figure 7 it can be seen that the result of ray tracing.Effectively
Intensity shows that the sine between 0 and 21 degree is multiplied by the expected profile of Gauss.Judge from the chromatic value of the light, 21 and 24 degree it
Between peak be due to the too high intensity from reflecting surface, it may be possible to due to the defect of Surface interpolating in LightTools.Color
Angle value is on two straight lines.Value between 0 and 1 degree is somewhat too high.This part light comes from the wide-angle from LED, and this is to survey
It is high by one during amount result shows some scramblings, and especially the y chromatic values in measurement result are than least square fitting
The place of point.The colourity of the light launched is completely contained in a MacAdam's ellipse (MacAdams ellipse), so
Aberration is not obvious for human eye.
Analog results of Fig. 7 by using LightTools in the case of two sections of collimaters.Here source has 0.01mm's
Width.Bar chart shows active strength.Open and closed dotted line shows x and y chromatic values.
5. conclusion
A kind of new method based on inverse method for designing is had been incorporated into for reducing the CoA changes in LED luminescent systems.Lead
Two equation groups of the ODE of coupling are gone out.Some characteristics of method for solving are discussed in Section 3, and the 4th
The numerical problem being integrated to ODE systems is solved in section.It has been directed to the LED spotlight with TIR collimaters and has realized this
Method.Followed the trail of by the Monte Carlo ray in ray tracking soft LightTools and test collimation for point source
Device.Experimental result shows that CoA changes are reduced in the limit value of visually-perceptible (Fig. 7).
Herein it has been shown that by using inverse approach, for spot light, the angle blend of colors in far field is possible
's.Similarly, it has been shown that by mixing only two from the light source light of the different angles of departure, all colours change may be removed
Change.
In research from now on, we are want to include the finite size of light source in method.LED is generally straight with millimeter
Footpath, this is excessive compared with being considered as the optical dimensions of point source.Additionally, we are interested to be directed to N>The solution of 2 (19)
Method.This will allow the calculating of the transmission function of the collimater for wider range, and allow TIR collimaters more to design
The free degree.Finally, the equivalent method for the LED of the xy colorimetric properties with curve will be interested.
Bibliography
1.‘ORA Light Tools product website’.http://www.opticalres.com/lt/
Ltprodds_f.html.Accessed February 27,2012.
2.Bortz, J.and N.Shatz:2006, ' Generalized functional method of
nonimaging optical design’.In:R.Winston&P.Benitez(ed.):Nonimaging Optics and Efficient Illumination Systems III, Vol.6338ofProc.SPIE.
3.Bortz, J.and N.Shatz:2010, ' Mathematical relationships between the
Gen-eralized functional, edge-ray and sms design methods ' .In:R.Winston&
J.M.Gordon(ed.):Nonimaging Optics:Efficient Design for Illumination and Solar Concentration, Vol.7785 ofNonimaging Optics:Efficient Design for Illumination and Solar Concentration.
4.Haitz, R.and Y.T.Jeffrey:2011, ' Solid-state lighting:’The case’
10years after and future prospects’.Physica Status Solidi A208,17-29.
5.Heath, M.:2005,Scientific Computing.An Introductory Survey.McGraw
Hill, 2nd edition.
6.Maes, M.:1997, ' Mathematical methods for reflector design '
.Ph.D.thesis, University of Amsterdam.
7.Maes, M.and A.Janssen:1991, ' A Note on cylindrical reflector design '
.Optik88,177-181.
8.Malacara, D.:2002,Color Vision and Colorimetry.Theory and Applications.SPIE Press.
9.Mueller, G.O., ' Luminescent ceramic for a light emitting device '
.United States patent US7361938B2.
10.Schubert, F.:2006,Light-Emitting Diodes.Cambridge University Press,
2nd edition.
11.Wang, K., D.Wu, F.Chen, Z.Liu, X.Luo, and S.Liu:2010, ' Angular color
uniformity enhancement of white light-emitting diodes integrated with
freeform lenses’.Optics Letters35,1860-1862.
12.Weisstein, E., ' Full Width at Half Maximum ' .http://
mathworld.wolfram.com/FullWidthatHalfMaximum.html.
13.Weisstein, E., ' Gaussian Function ' .http://mathworld.wolfram.com/
GaussianFunction.html.
14.Wu, H., N.Narendran, Y.Gu, and A.Bierman:2007, ' Improving the Per-
formance of Mixed-Color White LED Systems by Using Scattered Photon
Extraction Technique’.In:Proc.SPIE, Vol.6669ofProc.SPIE.
15.Wyszecki, G.and W.Stiles:2000,Color Science.Concepts and Methods, Quantitative Data and Formulae.John Wiley and Sons, INC, 2nd edition.